Everyday AI Podcast – An AI and ChatGPT Podcast - EP 553: Does AI Cause Brain Rot? What MIT's Viral Research Got Wrong

Episode Date: June 24, 2025

Does AI cause brain rot? If you read media coverage of MIT's recent viral study, you'd think yes. But that study completely missed the point of AI. Join us to figure out the truth behind the study, and how you should be using AI to make sure it's more than brain rot. Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion: Thoughts on this? Join the convo.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:MIT Study on AI Brain RotFlaws in MIT's Viral AI ResearchMedia Sensationalism on ChatGPTProper AI Usage and Brain HealthCognitive Debt Hypothesis and AIEffective AI Tools for Critical ThinkingAI's Role in Skill AugmentationAI in Education: Microsoft & Harvard StudiesTimestamps:00:00 "Everyday AI: Podcast & Newsletter"07:22 Media Misunderstands AI Research11:00 Testing AI with Random Participants13:43 "Students' Essays Rely on AI"16:38 Critique of MIT Study19:49 AI Use: Avoid Skill Erosion23:15 AI Enhances Learning and Decision-Making27:02 "AI: Tool or Shortcut?"28:42 Augmented Intelligence Over AI Dependency32:22 "No-Code AI for Business Growth"Keywords:AI brain rot, MIT study, ChatGPT brain activity, cognitive debt, memory failure, brainpower, large language models, ChatGPT essays, formulaic writing, AI brain impact, neural connectivity, creativity focus, thought partner, AI critical thinking, cognitive outcomes, AI study findings, ChatGPT users, bias in AI writing, educational outcomes, personalized AI tutor, AI and intelligence, AI shortcuts, brain augmentation, consultant AI use, interactive AI.Send Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info)

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Does AI cause brain rot?

If you've been paying attention to social media the past few days, or if you read a lot of tech news, you're probably seeing that headline a lot. That's because a recent MIT study all but said, hey, the more you write with chat GPT, the less you actually retain. Your brain is just rotting. Is it true? Maybe. Is this another case of research gone bad in the AI space?

Abs of friggin' lutele. So I'm going to be going over that today on MIT's new viral study, breaking it down and actually showing you what they got wrong and why I think you should pay attention to that more than the actual. sensationalized headlines that have been driving the conversation around this. All right. Welcome to Everyday AI. What's going on, y'all?

My name's Jordan Wilson, and I'm the host of Everyday AI. This is your daily live stream podcast and free daily newsletter, helping everyday business leaders like you and me, not just keep up with AI, but how we can get ahead and use all this new information and all these new tools, these new research papers to grow our companies and our careers. It starts here with the unedited, unscripted podcast and livestripping. but if you really want to take it to the next level and be the smartest person in AI at your company or in your department, that happens on our website at your EverydayAI.com.

There, you can sign up for our free daily newsletter. We're going to be recapping the highlights from today's show, but also you can go listen to 550 plus back episodes of the podcast for free, watch the videos, read the newsletters, all on our website, learn from the world's leading experts. So if you do want the AI news, make sure to go check in our newsletter for that. but let's get straight into it. It's hot take Tuesday, y'all, and I've got hot takes on this one. All right, but let me just tell you, stick around for the next 25-ish minutes. All right.

And I'm going to not just break down this viral study from MIT, but I might actually tear it apart and tell you why this whole thing is pretty much useless and give you my take on what proper AI use can actually do to your brain, not just rot it as this study so boldly claims. This is a fun one. Let's get into it. All right. Live stream audience, you'll see what I got on screen, a podcast audience. Don't worry.

I got a couple screen grabs here from some research papers. Nothing crazy. But here's essentially what this whole MIT study is about. the science itself is sound right very unlike the recent uh apple quote unquote research paper that uh i tore apart as did just about everyone else on the internet who understands AI this one's different right Apple's paper i call it weaponized marketing this viral paper from MIT the science is out, right? There's, there's no doubt that their findings are actually legitimate. But the premise of this study is just face palm. Like, come on. All right. So here's what their study actually

concluded. Right. So this new MIT study found that using chat GPT to essentially copy and paste and right things led to lower brain activity and memory failure. So yeah. Duh. Why, like, why did MIT, the MIT, the Massachusetts Institute of Technology, waste. How many pages was this thing? why do they waste a couple hundred digital 206 digital pages on a foregone conclusion right this is like you're like oh breaking news science confirms watching peloton ads doesn't burn calories yeah right right oh science confirms watching someone else exercise doesn't make you fit or, you know, watching someone else play the piano doesn't help you improve your ability to play it. Like, duh.

Why? Who approved this? Yes, the science is sound. I'm going to get into that. But the conclusions were predetermined, right? Of course, if you are using chat GPT to blindly copy and paste and write a research paper, Of course you're not going to retain any of the information. Right.

That's like saying like, oh, someone who went on Wikipedia and copy and pasted the information and submitted it didn't retain any of the information. Duh. Who approved this study? And I'm wondering, is it, is it fine if I ramp this up a bit as I sip on my coffee? It's not to take Tuesday. who approved this? Who, like, like seriously, this is this, it's, it's a growing trend of just mind-boggingly, like, mind-bogglingly useless AI studies. So I didn't know, like, do we actually need a study to confirm something that's a foregone conclusion, right?

And I don't know if it's, if it's, you know, researchers try to make a name- for themselves because I don't know. Maybe like researchers know right now that if you put out a study like this, number one, most journalists aren't going to read it. I'm a former journalist so I can say that. I can trash the media. I spent seven years as a journalist, right? Number one, journalists aren't going to read this whole thing.

Aside from a select few people in the media who really understand AI, most people don't. right all they're going to see is they're going to look at the um the beginning of this uh paper right the abstract they're going to read uh the abstract the summer results and the very conclusion of this so they're going to read about five percent of this paper and then they're going to slap the most sensationalized headline they can because all they're trying to do is get clicks right so now even though the MIT study uh you know didn't say chat gptt rot's your brain right? The actual name of the paper was your brain on chat. GBT accumulation of cognitive debt when using an AI assistant for essay writing tasks.

That's not going to sell clicks online. You know what does? ChatGPT gives you brain rot. So that's what like the entire media and social media has been running with. But this study is useless, right? New study confirms if you drop inexpensive glass. vase from the third floor under the ground, under the concrete, it's going to shatter.

So don't buy the vase, right? This, this vase is, uh, it's, it's breakable. Yeah, yeah, duh. I don't know. Why do we need this? All right. Let's get into more about this actual study, what it showed, how it was set up. Like I said, the research was sound.

The premise was, my gosh, like, I am literally at a loss for words. Again, my second straight hot take Tuesday going over a research paper, and I'm literally at a loss for words. Does anyone read these things before they get them approved? Or do we really need a research paper from MIT that says, you know, if you watch someone exercise, you don't burn calories? That's literally the closest equivalent I can make for this.

Anyways, the study set up an experiment designed to measure the cognitive cost of different AI tools in the study. Only 50 also, only 54 participants. Come on. So essentially 18 different people were split into three groups, 54 participants total. And there was essentially three different groups. One group used primarily chat chitb to write their essays. One was allowed to use Google to do research and then write their essays manually. And then the other ones, it was the brain-only group.

So they use a 32 electrode EEG to literally measure brain connectivity and track cognitive effort. And the core goal was to see if offloading writing to AI created a cognitive depth. So the EEG results clearly showed that using chat chitbt required the least amount of brain power. Really? Okay. Yeah. No one could have seen that coming except literally everyone. ChatGPT users had up to a 55% weaker neural connectivity than the brain-only group. And researchers showed the lowest engagement across frequencies tied to creativity, focus, and working memory. Also, human graders called the resulting AI-assisted essays, soulless, formulaic, and lacking personal insight. Yeah, that's exactly how large language models right by default, especially for people

that don't know what they're doing, right? It seems like for this study, they just, you know, recruited, you know, kind of random adults in the Boston area, right? There was no, it's not like these were a certain segment of the population. They were just adults, which, the ages, I believe it was 18 to 39. And yeah, like, of course, if someone just uses chat GBT or any large language model, and FYI, when I'm saying chat GBT in here, this is just what the researchers, you know, from MIT decided to, to measure, but swap out any name, right, Gemini, Claw, co-pilot, you know, lava, whatever. It's, it's, it's going to be the same.

So this isn't a knock on chat GBT. Right. This is just researchers are saying, hey, when you blindly write from a large language model, you're going to get essays that are soulless formulaic and lacking personal insights, which are exactly what large language models are trained to do by default. Right. They're not supposed to have personal insight. They're supposed to be extremely formulaic because it is a large language model. You can't get more formulaic than a large language model. That is literally what it is. And soulless, I hope.

Because if those Chad GPT essays had any soul, there would be a problem. That means, oh, we got H.E.I or, you know, artificial general intelligence or artificial superintelligence on our hands. Also, live stream audience, do you got any good takes on this? If you do, leave them in the comments. Or if you have a comment that's different or a take that's different. or a take that's different from mine. I'd love to hear that too. I wanted to feature some of your comments in today's newsletter where we're going to be recapping today's episode.

So yeah, please do. What's your hot take Tuesday? Right. Go ahead, throw that even though I got it wrong, even though I got it wrong there on my little ticker. Let me fix that. Go ahead. Put the hashtag hot take Tuesday. And what's your take on this?

Is this study legit? Was it needed? Did it change your mind on anything? I don't know. So some more findings from this paper. The most dramatic finding was that chat Chapti users couldn't remember what they had just written. I wonder why.

It's because they didn't write it. So the study showed that an incredible number of percentage, 83.3% of chat Chbt users failed to correctly quote their own finished essays because they don't read it. Right. I can only like imagine, now or professor or teacher being like, yeah, I feel this, right? This is college, right? Like, don't get me started because I will go for like an hour on how broken the higher education

system is in the U.S. But hey, if you if you still want to, you know, do these essays in college, the only way you should grade it, if you still want to do a 20 page paper, which is useless, every single student. I don't care if if you think they're the cleanest cut, student teachers out there, every single student is using AI tools to write their essay. So number one, we should probably stop using that as a measure of intelligence in the higher education system. Duh. Number two, if you must, you know, ask this student five questions in person, right, live about

their own essay. And one of two things will happen. Either every single student will get it wrong or probably one thing will happen, then this will happen. If you do that without telling them, all five will get it wrong. Or sorry, every single student will not be able to recall hardly anything in there because they didn't come up with it themselves. That's the first thing that would happen. The second thing that would happen is if you told them that you were doing it, they would probably at least read what they wrote.

I've talked to so many students. Students don't even read what they quote unquote write with chat GPT, right? They skim it. They're like, all right, this looks good. They copy and paste it. And that's it. So it's not shocking that in this study that 83% of chat GPT users fail to correctly quote their own finished essays. Also, this represents a complete memory failure compared to the only 11% in the other groups.

Yeah, that's about an eight times higher rate of cognitive debt or not being able to recall it. And that's what has led to all of these brain, AI, brain, brain. wrought articles and social media posts that have been flooding the airwaves, so to speak. All right, let me pause real quick in here from our sponsors. This podcast is supported by Google. Hey, everyone. David here, one of the product leads for Google Gemini. Check out VO3, our state-of-the-art AI video generation model in the Gemini app, which lets you create high-quality, eight-second videos with native audio generation. Try it with the Google AI Pro Plan or get the highest access with the Ultra Plan.

Sign up at jemini.com to get started and show us what you create. So there's obviously way better studies out there that have been published by reputable organizations that shed light on this topic in a much more and just an all-encompassing way. Like I said, this study from MIT is useless. I would say it's not worth the paper it's printed on, but hopefully no one printed out this 206 page essay or this 206 page study. But let's look at a couple. So, but before we do, a couple more things on what I'm saying is this tool swap twist that talked about this cognitive debt. debt. So the final session of this, so there was multiple sessions in this study, the groups switch.

So remember when I said previously, there was these three different groups. There was a chat GPT user group. There was a, oh, you could Google things, but you still had to write them yourself. And then there was those that used their brain only. So the final session is where these groups swapped tools or groups. And that was what ultimately confirmed the cognitive depth, that hypothesis. So when AI was taken away, users showed weaker brain networks and still couldn't recall their work. And they even reused specific AI phrases themselves, showing an internalized vocabulary and

style bias. In contrast, brain only users given AI for the first time used it strategically. So essentially there was this swap, right? So the group that first used Chad GPT and then their kind of brains only perform much worse than the group that first used their brains only and then Chad Chhabit. So essentially the first group, they kind of had this bias that when they were they had to use their brains, they were starting to parrot different talking points that they learned or that they wrote about, quote unquote, wrote about in their paper. So that's kind of what the, what the study showed. But like I said, this, this, this study is just irresponsible, right? We didn't need it at all, right?

To test a lazy, you know, copy and paste, you know, these users use chat chbt to write something. The outcome was already predetermined. And of course, no one is going to retain this information. So, uh, I think that this study probably before they started, they're like, yeah, this is to get some headline. This is going to grab some headlines. Let's let's go ahead and run this, right? It's this isn't realistic. This isn't realistic. Is it realistic that, you know, individuals for their companies, uh, you know, are, are doing this. Absolutely. Is this a problem? Yes. Like I said, the science is sound. And this is actually problematic because I've said this,

you know, I had an episode two marches ago. It's like March 20, 23 about how I truly believe AI makes most people dumb, right? The more you use it, the less you ultimately are using your brain. Because as humans, we become lazy and we just are looking for that big red, easy AI button that just spits things out for us, right? And that's why I try to do as best as I can warn everyone all the time as that's the way that you should not be using AI. You don't want to use it as an extension of your current skill set because then all that's going to happen as is evidence in this MIT research paper. All you're going to do then is erode your skills in that actual field. Right.

So that's why I really try and teach people the right way to use AI, which is more as a strategist, a consultant, a thought partner, you know, a brainstorming buddy, what have you versus an output machine, which is what most people use it as, because guess what, right? Whatever it is that you're using AI for, think, if you're using it to write blog posts, if you're using it to build, I don't know, financial reports, whatever it is that you're probably using it to do the bulk of what your job requires, your actual skills in those areas are plummeting. So is there a brain rot component?

to using AI? Absolutely. Will AI and can AI make you dumb if you are just blindly handing off your task to it? Yes. Absolutely. Right. So again, this study, the science is sound. It's just an absolutely useless study. So I think the real problem here is the workflow and not the technology. If they were smart, if the researchers were to smart, they would have set this up in a way that made sense, right? I would have set this up in a way of, okay, if we use brain only, so no internet, go ahead and write something. Then you use Google research only, and then you go write something then use a large language model and write it manually. Because guess what would have happened in that case? I think then we would have had some actual new science,

that was helpful for the future of humanity, for businesses, for students, for learners. That would tell us something because my guess and hypothesis would be the latter group or the last group that would use an LLM to study would be able to write the best information. That's because, you know, when used correctly, large language models are the best learning partner, right? Imagine, yeah, if you still did, oh, my gosh, right? The horror of having to write something on your own. But imagine being able to use chat chb t to do that versus traditional Google search versus no search. That would have been a study worth reading. It would have been a study worth the headlines. But hey, I guess however you have to do it to get your paper out there,

I don't know. I guess props to them for playing the press. It's not hard these days. Sorry to my old friends in journalism still. So let's look at some actual research papers that proved that point. So there was a great one from Microsoft. I thought I had a screenshot of it here. Oh, there we go. So this is using research as a thought partner from Microsoft. And this research proved that when you use AI versus when you don't, when you do use AI, you have better outcomes than when you just use your own brain. So essentially, what I just show. showed, this study showed that AI used as a thought partner enhances a user's critical thinking and decision making.

Right? Why? Like, that's a sound study, MIT. Another study that I thought was really good here. This one, it was a 2025 study from Harvard, very similar to the study for Microsoft. So this 2025 study from Harvard show. that interactive AI can dramatically improve educational outcomes when used correctly. So this one used a randomized trial, and it found that an AI tutor personalized for

the user doubled students learning gains over traditional teaching. My gosh, there's some research that makes sense. So when you think, oh, AI's brain rot, yeah, if you just blindly have it, do you work and you don't read it and you don't put your own brain into it. Of course. Of course. Right. That's like saying, you know, if you paid someone, you know, to write your essay and you didn't read it, you're not going to retain any of the information, right? But if you hire a super smart tutor who sits with you and works with you one on one according to your own learning abilities and then you write it yourself, that's fantastic. Right. And that's what AI gives us.

That's why like, man, this brain rod study, I'm like, come on. Number one, stupid study. I've said that enough. Number two, media got played. Number three, what I think the problem here is, this is casting a shadow on AI that it doesn't deserve, right? You should go poke. Hey, researchers at MIT everywhere else. Yeah, drag AI.

I don't care. It's not like I'm like sitting here, oh, defending AI. No, like go poke AI and go, go. create sound scientific studies that improve AI, right, that showcase its biases, that showcase its, that showcase its, you know, reflection of the bad parts of society, you know, make systems through your research safer and more secure. Look at jail breaks and in things like that, things that truly make AI safer and better for everyone versus, you know, printing up a study that is just parroting a foregone conclusion.

So there's this thought out there that, you know, AI is a shortcut to stupidity. And I think that that narrative is wrong, right? Yes, if you sit there, if you sit there, and if you are just using AI to do your job, you're going to get dumb. Like, there's no way around. it.

If you are using it as a consultant, right? That's what we've always taught in our, you know, prime prompt polish course, you know, that gives you not just proper engineering, proper prompt engineering. That's the way we should all be using AI, right? You start by giving it your best, right? You give it your insights, your thoughts, your beliefs, your data, your outline, right? You give it the best of you and have it tear that apart and then you build with it. So if you're just using AI as a shortcut to stupidity, you know, play stupid games, win stupid prizes.

But we, y'all, we have the most powerful technology known to mankind. And the majority of people are using it as a shortcut to stupidity. but you, dear listener, can use it differently. Don't use it just for an output. Use it to build those skills, right? Let's say you're a seven out of ten when it comes to writing. If in this study, if you just blindly hand over your future writing abilities or your future writing tasks or your future writing projects and deliverables to chat,

that skill set is going to go down. next week or next month, you're going to be a six out of ten. The week or month after that, you're going to be a five out of ten. Pretty soon, you're not going to be able to string more than two coherent sentences together. If you just over-rely on AI to do a core skill set, instead, have it work with you every step of the way. So next week, you're a seven out of ten. The week after that, you're an eight out of ten. Then you're a nine out of ten.

That's how we should be using AI to augment our current abilities, not just to, just create as many shortcuts because I think the real risk isn't brain rot. It's being left behind by your competitors who are mastering this augmentation, right? That's where sometimes even the word artificial intelligence, I just like using augmented intelligence because we can't just hit the off switch on our own intelligence, which is what this MIT study that's, that's one of their three groups, they literally just had them power off their brains, right? And then say, oh, their brains didn't work. Well, yeah, you had them powered off and you had them hand off the whole assignment to AI. So this study proved nothing. But what I want you to do is you have

to start separating the tool from the workflow. All right. So what that means is don't just use AI for outputs. Use it at every step of the work. way. Human work is not going away anytime soon. It might look very different in a couple of months than it does today, right, especially when we talk about agentic AI. But recent evidence shows that using AI as an interactive thought partner and that approach to augment your current intelligence actually improves your cognitive outcomes. So you can choose brain rot if you'd like and you can power off your brain and you can give every single assignment deliverable work task to AI.

You can do that. And the AI systems are going to continue to get better and better. What does that do for you? Yeah, you'll probably get better at using AI and that's important. But when we talk about the future of AI, agentic AI, you still need those skill sets. Right. When I talk about expertise in the loop, because I hate human in the loop, where are you going to be sharpening that expertise if you're just constantly handing it off blindly to AI. That's brain rot.

Right? Use it as a coach, not a crutch. That's it. That's a rap. Hey, I said this one would only be about 25 to 30 minutes. About dang time, right? So, hey, here's the quick recap. Does AI cause brain rot? It can. Right? If you are looking, for shortcuts. If you look at AI and you say, I want to automate literally everything I can and turn my brain off, AI will cause brain rot. But that is a decision that the human makes,

not the default setting. The default setting for AI is we have the most powerful technology. ever at our fingertips. And whether it is brain rot or you becoming the smartest person in the room, that decision is up to you. And I hope you make the latter and not the former. Thanks for tuning in.

If you missed anything in this one, go to your everyday AI.com. Make sure to sign up for that free daily newsletter. If this was helpful, tell someone about it. If you're listening on the podcast, appreciate it, Please make sure to subscribe to the show and follow it. Share it with a friend if you could. You know what?

If you have shared this with someone, let me know. I always like reading those stories. I super appreciate that. It warms my heart. All right. So thank you for tuning in. Make sure to join us tomorrow. Our new Put AI to Work on Wednesday series,

we're going to be talking about no code interactive AI, how to use chat GPT and Google Canvas modes, as well as Claude's artifact mode for business growth. So it's going to be a good one. Make sure to tune in. Thank you for tuning in. Hope to see you back tomorrow for that one.

And every day for more everyday AI. Thanks y'all. Meet Firefly AI assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps,

including Photoshop, Premiere Express, and more in one conversational interface. You direct the output. come while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI. Thanks for joining us.

If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next. time.